Heatmap

What a heatmap shows

A heatmap shows how a page is used as a field of colour. Areas with much activity appear warm, areas with little appear cold. Three types are common. The click map shows where clicks happened, including on places that are not buttons at all. The scroll map shows how far visitors get down the page and therefore which part is seen at all. The attention map estimates from mouse movement and dwell time where the eye went, which stays an estimate rather than a measurement.

What it cannot show

A heatmap says nothing about intent, satisfaction or reasons. A warm area on a button does not mean the button is good. It can also mean visitors take it for something else and click wrongly. A cold area likewise does not mean the content is unimportant: it may simply sit below the point where most visitors stop scrolling.

That is why combining it with session recordings works so well. The heatmap finds the place, the recording explains what happens there.

What to watch when collecting

• Enough data: with a few hundred visitors a blotchy pattern appears that is shaped by individuals. It becomes telling only with a few thousand.
• Separate views: mobile and desktop have to be judged apart. A combined map shows an average that no device experiences.
• Separate campaign traffic: visitors from an ad behave differently than regular customers. Mixing both in one map hides both.
• No personal content: form inputs should be masked in recordings, otherwise data protection problems arise.

What follows in practice

Three findings repeat often. First, the buying incentive often sits below the area most visitors reach. The answer is then less content above rather than more content overall. Second, elements near advertising banners or discount strips get overlooked because they look like ads. Third, people click places that are not clickable, such as product names in a list. That produces a concrete change instead of a design debate.

Frequent questions

• Does a heatmap need consent? If it processes personal data, yes. Without consent, aggregated map views can be used.
• Does it replace A/B testing? No, it finds starting points. Proof comes from a test.
• How often should I look? After every larger change, and then at wider intervals.